From: Michael Zimmermann <sigmaepsilon92@gmail.com>
To: "Gao, Liming" <liming.gao@intel.com>
Cc: Ard Biesheuvel <ard.biesheuvel@linaro.org>,
"edk2-devel@lists.01.org" <edk2-devel@ml01.01.org>,
Star Zeng <star.zeng@intel.com>, Feng Tian <feng.tian@intel.com>
Subject: Re: correct way to reserve memory from PrePi?
Date: Thu, 15 Dec 2016 06:02:02 +0100 [thread overview]
Message-ID: <CAN9vWDJi6LsxhW61EpwFgot9kpPOCorAEr2v_f3xtMw=Q3Q-hw@mail.gmail.com> (raw)
In-Reply-To: <4A89E2EF3DFEDB4C8BFDE51014F606A14D6A55EE@SHSMSX104.ccr.corp.intel.com>
I do not want to allocate the full memory range, just small parts of
it. I have one or two ranges(depending on how the system config
reports it) for all ram. e.g a 2gb device usually uses either
0x0000000-0x80000000 or 0x00000000-0x40000000 and
0x40000000-0x40000000.
So I digged even further and it looks like upon boot, the highest
available memory resource descriptor is selected and allocated for the
DxeCore.
So if I have one memory resource only, the whole DRAM will be
allocated by/for DxeCore. Is that really the intended behavior?
Because this way I'm unable to reserve any RAM from the highest
resource descriptor. The reason why it works when I have split the
DRAM in two descriptors is that DxeCore will only allocate the upper
half and I'm usually trying to reserve a few low memory ranges - which
isn't always the case.
Thanks
Michael
On Thu, Dec 15, 2016 at 5:55 AM, Gao, Liming <liming.gao@intel.com> wrote:
> Michael:
> I understand your usage that BuildResourceDescriptorHob adds system memory range, BuildMemoryAllocationHob allocate the full memory range as reserved memory. Then, you expect they can be shown in EFI memory map. Right?
>
> BuildResourceDescriptorHob() with (start,size) 0x80000000,0x80000000, then BuildMemoryAllocationHob() with (start,size) 0x80000000,0x80000000. It doesn't work.
> Two BuildResourceDescriptorHob() with (start,size) 0x80000000,0x40000000 and 0xc0000000,0x40000000, then two BuildMemoryAllocationHob(). It does work. Right?
>
> Thanks
> Liming
>> -----Original Message-----
>> From: edk2-devel [mailto:edk2-devel-bounces@lists.01.org] On Behalf Of
>> Michael Zimmermann
>> Sent: Thursday, December 15, 2016 1:57 AM
>> To: Ard Biesheuvel <ard.biesheuvel@linaro.org>
>> Cc: edk2-devel@lists.01.org <edk2-devel@ml01.01.org>
>> Subject: Re: [edk2] correct way to reserve memory from PrePi?
>>
>> I've enabled GCD debugging and apparently it doesn't accept the allocation:
>> GCD:AllocateMemorySpace(Base=0000000090000000,Length=0000000010000
>> 000)
>> GcdAllocateType = AtAddress
>> GcdMemoryType = SystemMem
>> Alignment = 0000000000000001
>> ImageHandle = FDE28F90
>> DeviceHandle = 0
>> CoreAllocateSpaceCheckEntry:982 handle=FDE28F90
>> CoreAllocateSpace:1130
>> Status = Not Found
>> GCDMemType Range Capabilities
>> Attributes
>> ========== =================================
>> ================
>> ================
>> NonExist 0000000000000000-000000007FFFFFFF 0000000000000000
>> 0000000000000000
>> SystemMem 0000000080000000-00000000FDFFFFFF 800000000000000E
>> 0000000000000000*
>> NonExist 00000000FE000000-00000000FE3FFFFF 0000000000000000
>> 0000000000000000
>> SystemMem 00000000FE400000-00000000FFFE5FFF 800000000000000E
>> 0000000000000000
>> SystemMem 00000000FFFE6000-00000000FFFEEFFF 800000000000000E
>> 0000000000000000*
>> SystemMem 00000000FFFEF000-00000000FFFFFFFF 800000000000000E
>> 0000000000000000
>>
>>
>> It fails in this line:
>> https://github.com/tianocore/edk2/blob/master/MdeModulePkg/Core/Dxe
>> /Gcd/Gcd.c#L982
>> In this example I've used(start,size) 0x80000000,0x80000000 for the
>> resource descriptors.
>> If I use 0x80000000,0x40000000 and 0xc0000000,0x40000000 everything
>> works just fine.
>>
>> Thanks
>> Michael
>>
>> On Wed, Dec 14, 2016 at 4:13 PM, Michael Zimmermann
>> <sigmaepsilon92@gmail.com> wrote:
>> > As far as I know the proper way is to create resource descriptors
>> > using BuildResourceDescriptorHob and then allocate reserved areas
>> > using BuildMemoryAllocationHob. This way I don't have any overlapping
>> > descriptors - I just allocated some memory very early.
>> >
>> > I ran many tests and it looks like all calls to
>> > BuildMemoryAllocationHob get ignored if my dram hobs look like this:
>> > 0x00000000 - 0x40000000
>> >
>> > If I split this range into two Hob's like this everything seems to
>> > work just fine:
>> > 0x00000000 - 0x20000000
>> > 0x20000000 - 0x20000000
>> >
>> > I took a look at other platforms like Juno and they add big dram Hob's
>> > (2GB and 6GB) too so why is this a problem?
>> >
>> > Thanks
>> > Michael
>> >
>> > On Wed, Dec 14, 2016 at 11:21 AM, Ard Biesheuvel
>> > <ard.biesheuvel@linaro.org> wrote:
>> >> On 14 December 2016 at 10:02, Michael Zimmermann
>> >> <sigmaepsilon92@gmail.com> wrote:
>> >>> I tried both BuildResourceDescriptorHob and BuildMemoryAllocationHob
>> >>> but apparently they don't have any effect.
>> >>> When I look at the output of the shell's memmap command there aren't
>> >>> any reserved/unavailable pages.
>> >>>
>> >>> Furthermore, when using AllocatePages with one of the physical
>> >>> addresses which I've reserved it succeeds which means that it's not
>> >>> just a problem of how the memmap command works.
>> >>>
>> >>> Am I doing something wrong or is this a bug?
>> >>>
>> >>
>> >> I think you need to ensure that they don't overlap existing resource
>> >> descriptors: if you declare a region as reserved, you should not
>> >> declare it as memory first
>> _______________________________________________
>> edk2-devel mailing list
>> edk2-devel@lists.01.org
>> https://lists.01.org/mailman/listinfo/edk2-devel
> _______________________________________________
> edk2-devel mailing list
> edk2-devel@lists.01.org
> https://lists.01.org/mailman/listinfo/edk2-devel
next prev parent reply other threads:[~2016-12-15 5:02 UTC|newest]
Thread overview: 8+ messages / expand[flat|nested] mbox.gz Atom feed top
2016-12-14 10:02 correct way to reserve memory from PrePi? Michael Zimmermann
2016-12-14 10:21 ` Ard Biesheuvel
2016-12-14 15:13 ` Michael Zimmermann
2016-12-14 17:57 ` Michael Zimmermann
2016-12-15 4:55 ` Gao, Liming
2016-12-15 5:02 ` Michael Zimmermann [this message]
2016-12-15 5:12 ` Gao, Liming
2016-12-15 5:27 ` Michael Zimmermann
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-list from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to='CAN9vWDJi6LsxhW61EpwFgot9kpPOCorAEr2v_f3xtMw=Q3Q-hw@mail.gmail.com' \
--to=devel@edk2.groups.io \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox